Spice provides an open‑source SQL runtime that federates queries across operational databases, data lakes, and warehouses without moving data, then accelerates hot tables with local engines for sub‑second response times. It extends standard SQL with hybrid search functions and built‑in LLM inference, enabling context‑aware retrieval and AI‑driven workflows while maintaining enterprise governance and cost efficiency.
Funding
$13.5M raised to dateRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
1OMFounders
Product
Problem
Enterprises often struggle with siloed data across operational databases, data lakes, and warehouses, requiring complex ETL pipelines and multiple query engines to achieve low-latency access for analytics, search, and AI workloads. This fragmentation leads to high latency, elevated storage costs, and governance challenges when integrating large language models with enterprise data.
Solution
Spice delivers a unified, open‑source SQL runtime that federates queries across heterogeneous data sources without moving data, then accelerates frequently accessed tables using local engines such as DuckDB or SQLite for sub‑second response times. The platform extends standard SQL with hybrid search functions that combine vector similarity, full‑text, and relational filters, enabling context‑aware retrieval at scale. Built‑in LLM inference functions let developers call hosted or on‑premise models directly from SQL, supporting summarization, classification, and RAG pipelines without external glue code. Spice can be deployed as a sidecar, microservice, or fully managed cloud service, providing consistent performance and governance from edge to cloud. Integrated observability, role‑based access control, and secure AI sandboxes ensure data compliance while delivering up to 100× faster queries and significant cost savings on data lakehouse workloads.
Target Audience
Spice targets data engineering, analytics, and AI teams in medium to large enterprises that need low‑latency, governed access to federated data for real‑time applications, dashboards, and AI agents.
Features
- Federated SQL engine that queries operational databases, data lakes, and warehouses in a single layer with zero ETL
- Automatic materialization and indexing of hot tables using DuckDB, SQLite, or similar engines for millisecond‑scale access
- Hybrid search capabilities that merge vector embeddings, keyword BM25, and relational filters via standard SQL functions
- AI() SQL function for inline LLM calls (OpenAI, Anthropic, Bedrock, or custom endpoints) supporting summarization, translation, and classification
- Natural‑language to SQL generation for rapid query creation and RAG workflows
- Secure AI sandboxing with fine‑grained, temporary data scopes and full audit logging
- Built‑in metrics, distributed tracing, and role‑based access controls for enterprise governance
- Over 30 prebuilt connectors for common sources (Databricks, S3, MySQL, PostgreSQL, Snowflake, etc.) and support for Arrow Flight, JDBC/ODBC